Must read for anyone running regressions of any kind and using R -- that is, every scientist out there.

Models as Prediction Machines: How to Convert Confusing Coefficients Into Clear Quantities - Julia M. Rohrer, Vincent Arel-Bundock, 2026 Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear model...

Good news everyone 🥳 Our (w @vincentab.bsky.social) primer on models as prediction machines (with the marginaleffects package) is finally officially published!>

journals.sagepub.com/doi/10.1177/...

Context-dependent violations of rational choice in honeybees (Apis mellifera) and gray jays (Perisoreus canadensis) - Behavioral Ecology and Sociobiology Contrary to the theory of rational choice, adding an alternative to a set of available options often affects people's judgement of the preexisting options. Here, we show that honeybees (Apis mellifera...

There are several papers on the attraction effect -- which is a sort of generalized Monty Hall -- run on animals. The effect was found on bees, on gray jays (here: link.springer.com/article/10.1...), and on *slime mold* that doesn't even have a brain (pubmed.ncbi.nlm.nih.gov/20702460/).

here goes the tagline for the paper I use in presentations

Humans do pick it up if given time & incentives. We have a paper on the attraction effect -- a sort of generalized Monty Hall -- and this is the dynamics -- "up" means stronger effect (ie staying), zero means doing what's best given the stimuli (ie switching).

pubsonline.informs.org/doi/10.1287/...

European kidney exchange is making progress:)
#ECONSKY
marketdesigner.blogspot.com/2026/04/lund...

There are others outside of the grunge era. You have No surprises / karma police. You have pop songs like Mr Jones and even more pop like All Star. Basket Case from the punk scene. And all are very different.
But I agree smells like teen spirit is the only correct answer for No.1

Alive

Bernhard Dalheimer and I were interviewed on the Eat This podcast about our recent working paper on the geopolitics of food and agriculture. You can listen here: eatthispodcast.com/geopolitics/

Or the Italian version

1. Giovane promessa
2. Solito stronzo
3. Venerato maestro

and three-way interaction subgroup analyses to boot

🙋 I am. OSF is über-bloated, really hard to find what I actually want to find from an OSF page.

Very useful work from Matti here.

Publishers extract enormous rents from the community and in exchange provide us with bloated websites.

Reducing their rent seeking is a large endeavor, but while we're on it at least we can streamline our user experience by ditching their bloat & frills.

Thanks!

PDF-Direct Firefox extension ('Bypass "fancy online reader" and just get the darn pdf') updated: addons.mozilla.org/en-US/firefo... & now supports

ACM
ACS
Cambridge University Press
Elsevier
IEEE
JSTOR
Nature
OUP
PLOS
PNAS
Royal Society Publishing
Sage
Science
Springer Nature
Taylor & Francis
Wiley

I’ll be leaving PCI in May to begin a new postdoc, which means my current position will soon be open!

You’ll find the job advert below (in French). Please feel free to share it within your networks, and don’t hesitate to get in touch if you’d like more information!

yes, but what I am talking about is not candidates coercing citizens at the voting booth. It's mafia leaders going around the neighbourhood before the election giving strict orders and providing credible threats of control.

making the ballot-taking public increases chances of control.

having a powerful social control by mafia hardened the Italian system making it converge to the simplest possible rules that can *in principle* and by default rule out most external controls, and do not rely on post-hoc controls.

agree that countermeasures exist, but those are second-order patches to a first-order problem: put all names on ONE ballot and let the voter cross one, & you don't need all this sending ballots by mail, ballot bins, controls, etc...

Impressive that the response of nearly all gov'ts but 🇫🇷 to "we have a chronic dependence on stuff that is controlled by a few bad guys & whose flow can stop at any moment for reasons out of our control" was "let's find another pusher" and "let's burn taxpayer money to increase our dependence".

the taking of ballots is publicly observable: this opens the way to control.

eg: tell specific voters to take only specific ballots, and claim that you'll observe them.

eg: ask people to take 2 ballots, post one, keep the other in their pocket, show it on exit.

20$/month buys you 6 prompts / 24h??? that's insanely expensive.

[I am on claude free tier, might upgrade at some point to use claude code]

Astonish quirk of the French system: N ballots, one per candidate/list, people take as many as they want, throw most away, put the chosen one in the box.

This is easily controllable by an organized external actor.

Why not have ONE ballot and people cross their choice, like in any other country?

Unfortunately Proton has no client for their ProtonDrive on Linux. This was the showstopper for me and I had to keep my Dropbox subscription for another year. Bugger

First box of books

The supply chain is getting ready:)
#econsky
marketdesigner.blogspot.com/2026/04/firs...

Who ya gonna call? the

🚨🇮🇹🤌Italian Food Police🤌🇮🇹🚨

Three recent papers examined reproducibility of a large sample of findings. Join this webinar to discuss them and explore where the findings converge and differ.

1. SCORE: www.nature.com/articles/s41...
2. I4R: www.nature.com/articles/s41...
3. Meta-Rep: royalsocietypublishing.org/rsos/article...

wait why, aren't there in other seas too?

I'm delighted that our paper, "Risk Matters Less When Options Are Apples-to-Oranges: The Translate-and-Accommodate Model" (Evan Weingarten, Yuval Rottenstreich, and George Wu) has just been accepted at Psychological Review.

papers.ssrn.com/sol3/papers....

1/N

Some years ago I got a similar email regarding then right-wing politician, now Italy's Foreign Minister, Crosetto.

They were *extremely* bad at it, though, as he's called Guido, not Paolo.

A mosaic chart titled "Roads and cycle paths." Two columns, bicycle and car, divided into rows by maintainer type (Private, Municipal, State). The car network dominates: Municipal roads account for 3 488 km, State roads 941 km, and Private roads 562 km. The bicycle network is far smaller: Municipal 446 km, State 47 km, Private 42 km. Source: NVDB, 2025.

#day3 of #30DayChartChallenge, Mosaic

code: github.com/gkaramanis/3...

#RStats #dataviz